Noise influence on context formation and lexical retrieval in speech comprehension

Context-based lexical predictions facilitate speech comprehension, but noise can disrupt these mechanisms by degrading context formation and word retrieval. This study investigated how noise affects the word predictability effect during comprehension through analysis of three event-related potentials: The N200, N400, and late positive component (LPC). Sixty participants listened to clear and noise-masked sentences ending with high- or low-predictability words for comprehension across two experiments. In Experiment 1, we masked sentence frames to degrade context while preserving final words. In Experiment 2, we masked final words to hinder retrieval while preserving context. Clear speech in Experiment 1 elicited a typical N400 predictability effect. Importantly, noisy contexts elicited both a frontal N200 predictability effect and an unattenuated early N400 effect, suggesting that enhanced sound-meaning processing maintains comprehension despite continuous noise. In Experiment 2, both clear and noisy final words produced N400 and posterior LPC predictability effects, indicating that reanalysis helps consolidate word representations affected by transient and unpredictable sentence-end noise. These findings highlight how context formation and word retrieval jointly shape prediction under adverse listening conditions. The N200 and LPC effects reveal mechanisms of early phonological analysis and late reanalysis that support comprehension despite challenging acoustics. Our results illuminate the distinct cognitive mechanisms that support comprehension resilience through context use and word processing, advancing theories of speech comprehension.